α 1 -Microglobulin Is Found Both in Blood and in Most Tissues

Abstract

SUMMARY In this study we demonstrate that, in addition to blood, _α1-microglobulin (_α1m) is present in most tissues, including liver, heart, eye, kidney, lung, pancreas, and skeletal muscle. Western blotting of perfused and homogenized rat tissue supernatants revealed _α1m in its free, monomeric form and in high molecular weight forms, corresponding to the complexes fibronectin-_α1m and _α1-inhibitor-3-_α1m, which have previously been identified in plasma. The liver also contained a series of _α1m isoforms with apparent molecular masses between 40 and 50 kD. These bands did not react with anti-inter-α-inhibitor antibodies, indicating that they do not represent the _α1m-bikunin precursor protein. Similarly, the heart contained a 45-kD _α1m band and the kidney a 50-kD _α1m band. None of these _α1m isoforms was present in plasma. Immunohistochemical analysis of human tissue demonstrated granular intracellular labeling of _α1m in hepatocytes and in the proximal epithelial cells of the kidney. In addition, _α1m immunoreactivity was detected in the interstitial connective tissue of heart and lung and in the adventitia of blood vessels as well as on cell surfaces of cardiocytes. _α1m mRNA was found in the liver and pancreas by polymerase chain reaction, suggesting that the protein found in other tissues is transported via the bloodstream from the production sites in liver and pancreas. The results of this study indicate that in addition to its role in plasma, _α1m may have important functions in the interstitium of several tissues.

_α1-Microglobulin (_α1m) is a yellow-brown 26-kD glycoprotein which was originally isolated from the urine of patients with renal defects (Ekström et al. 1975; Tejler et al. 1976; Ekström and Berggård 1977). _α1m is a member of the lipocalin superfamily, which now includes approximately 25 members (Flower 1996). Retinol binding protein (RBP), _α1-acid glycoprotein (orosomucoid), and prostaglandin D-synthase are other members of this family and, like _α1m, are predominantly extracellular proteins. A majority of the lipocalins bind and transport specific small hydrophobic substances, such as retinoids and steroids, in a hydrophobic pocket. This pocket is formed by eight antiparallel β-strands folded into a β-barrel, lined on the inside by hydrophobic residues that specify the ligand specificity of each lipocalin (Papiz et al. 1986; Cowan et al. 1990). The biological significance of the homology between α₁m and the other lipocalins is presently not known because the binding specificity of α₁m has not been identified.

α₁m has a strong tendency to form complexes with other proteins. In human and rat, approximately half of the plasma α₁m exists as high molecular weight forms, which are too large to escape from the circulation via glomerular filtration. Curiously, the α₁m complexes that have been found thus far in human plasma are not found in rat plasma, and vice versa. α₁m has been shown to be complexed with IgA, prothrombin, and albumin in humans (Tejler et al. 1976; Grubb et al. 1983; Berggård et al. 1997) and with fibronectin and the α₂-macroglobulin homologue α₁-inhibitor-3 in rats (Falkenberg et al. 1990,1994). The sites of formation and metabolism of these high molecular weight α₁m complexes are unknown, but the α-inhibitor-3-α₁m complex was shown to be rapidly cleared from the circulation (Falkenberg et al. 1995).

The gene for α₁m has been isolated and characterized (Diarra-Mehrpour et al. 1990; Vetr and Gebhard 1990) and encodes two tandemly arranged proteins, α₁m and bikunin (Kaumeyer et al. 1986). Bikunin is the light chain found in the plasma proteins inter-α-inhibitor, pre-α-inhibitor, and heavy chain 2-bikunin (Enghild et al. 1989). The bikunin subunit is responsible for the proteinase inhibitory activities of these proteins. Bikunin also apparently stimulates the growth of normal endothelial cells (McKeehan et al. 1986) and fibroblasts (Perry et al. 1994). Moreover, inter-α-inhibitor has been shown to be an important component for the formation of the extracellular matrix (Chen et al. 1992; Blom et al. 1995).

When synthesized, a precursor protein is translated in which α₁ m and bikunin are connected by a short peptide. The precursor is cleaved into the two mature proteins before secretion (Bratt et al. 1993,1994; Thøgersen and Enghild 1995). The co-expression of α₁m and bikunin is conserved in all species examined to date (from fish to human) (Hanley and Powell 1994; Leaver et al. 1994). This suggests that the two proteins are involved in common processes. However, thus far no functional relation has been shown between α₁m and bikunin in plasma. α₁m has been extensively studied in blood, but although different immunoregulatory properties have been reported for α₁m (for reviews see Åkerström and Lögdberg 1990; Åkerström 1992), the physiological function of the protein is still unclear.

α₁m is generally considered to be a plasma protein, but the result reported in this study reveals that α₁m is present in the extracellular matrix of many tissues. Moreover, α₁m is not generally synthesized in the tissues but mainly in the liver, and is probably transported to the tissues across the endothelial cell membrane by an unknown mechanism. In addition, both high and low molecular weight forms of α₁m are widely distributed in the matrix of tissues, suggesting a possible function in these interstitial tissue domains.

Materials and Methods

Materials

The general serine-proteinase inhibitor 3,4-dichloroisocoumarin (DCI) and the general cysteine-proteinase inhibitor N-[[[N[-[(-3-trans-carboxyl-oxiran-2-yl)-carbonyl]L-leucyl] amino]-butyl]-guanidine (E-64) were from Boehringer-Mannheim (Indianapolis, IN). Goat anti-rabbit IgG-alkaline phosphatase-conjugated antiserum, 1,10-phenanthroline bromochloroindolyl phosphate, and nitroblue tetrazolium were purchased from Sigma (St Louis, MO). Rabbit antisera against human α₁m were purchased from Accurate Chemical and Scientific (Westbury, NY). Rat plasma was obtained from Pel-Freeze Biologicals, (Rogers, AR). The rabbit antiserum against rat α₁m was prepared as described previously (Åkerström and Landin 1985). Human α₁m was purified from the urine of patients with tubular proteinuria as described (Åkerström et al. 1995). Rat α₁-inhibitor-3 was purified as reported earlier (Enghild et al. 1989). Sprague-Dawley rats were from Charles Rivers (Wilmington, MA). Human tissue samples for immunohistochemistry and PCR analysis were obtained at autopsy by the Department of Pathology, Duke University Medical Center (Durham, NC) and were histologically determined to be normal.

SDS-polyacrylamide Gel Electrophoresis

Aliquots of the supernatant from each of the homogenized tissue samples were subjected to SDS-PAGE under nonreducing conditions. This was performed in 5-15% linear gradient gels using the glycine-2-amino-2-methyl-1,3-propanediol-HCl system described by Bury (1981). Molecular mass standards were rat α₁-inhibitor-3 (180 kD) mixed with Bio-Rad (Hercules, CA) low molecular weight mass standard, visualized by staining with Coomassie Brilliant Blue.

Western Blot Analysis

Proteins were separated by SDS-PAGE and transferred to polyvinylidene diflouride membrane (PVDF) as described by Matsudaira (1987). All subsequent steps were performed on a rotating table at 25C. The membrane was equilibrated in 20 ml of 10 mM Tris-HCl, 0.15 mM NaCl, and 0.05% NP-40 (TSN buffer) containing 1% bovine albumin for 30 min before the addition of 10 μl α₁m antiserum (1:2000 dilution). After 1 hr the membrane was washed twice in TSN buffer and twice in 10 mM Tris-HCl, 10 mM NaCl, pH 7.5 (TS buffer) for 5 min each. Then 20 ml of TSN buffer containing 10 μl anti-rabbit IgG-alkaline phosphatase conjugate (1:2000 dilution) was allowed to react for 30 min before the membrane was washed as above. The substrates bromochloroindolyl phosphate and nitroblue tetrazolium were added and color was allowed to develop in the dark for up to 30 min. The reaction was stopped with 5 mM EDTA in TS buffer.

Perfusion of Rat Tissues

Sprague-Dawley rats were anesthetized by IP injection with pentobarbital. After sedation they were opened and one lung was removed. A blood sample was taken from the left ventricle of the heart, and the animal perfused via the same needle, using 500 ml of 0.9% normal saline. Perfusion of the lungs was performed by infusion of 100 ml of 0.9% saline through the right ventricle of the heart after cutting the abdominal aorta. The organs were removed, weighed, and homogenized in a high-speed homogenizer (Cole-Parmer Instruments; Niles, IL) in 100 mM NaCl, 50 mM (N-[2-hydroxyethyl]piperazine-N'-[2-ethanesulfonic acid] (HEPES), 100 mM 3,4-dichloroisocoumarin (DCI), 500 μM 1,10-phenanthroline, and 50 μM N-[[[N[-[(-3-trans-carboxyloxiran-2-yl)-carbonyl]l-leucyl]amino]-butyl]-guanidine (E-64) (10 ml/g). The homogenates were then centrifuged at 2500 rpm for 15 min and the supernatant was collected.

Immunohistochemistry

Human tissues were fixed in 10% formalin followed by routine processing and paraffin embedding. Sections were then labeled for antigen using an indirect immunoperoxidase method (Milde et al. 1989; Randell et al. 1991) with a biotinylated goat anti-rabbit IgG and streptavidin-horseradish peroxidase. To reduce background staining, the sections were incubated in 1% H₂O₂ in methanol to inactivate endogenous peroxidases and 10 mM borohydride to block aldehydes. Nonspecific binding was blocked by incubation with 5% normal goat serum (NGS), 5% milk, and 1% BSA in PBS. Primary and secondary antibody dilutions were determined empirically and made in PBS with 1% milk plus 1% BSA (milk was not included in the streptavidin solution). The slides were developed using diaminobenzidine (10 mg diaminobenzidine, 50 ml 0.05 M Tris-Cl, pH 7.6, 100 ml 3% H₂O₂) and counterstained with hematoxylin.

As a control, one serial section on each slide was incubated with preimmune rabbit serum or with anti-α₁m anti-bodies absorbed by immunosorbent affinity chromatography. The latter was done by immobilizing 1 mg of purified α₁m to 1 ml CNBr-activated Sepharose according to the manufacturer's instructions (Sigma). The affinity gel was then incubated for 24 hr with 1 ml of primary antibodies at two times the concentration used for immunolabeling. The supernatant was then used for immunohistochemistry.

Reverse Transcriptase PCR

The presence of mRNA in the tissue was examined by extracting RNA essentially as described previously (Chomczynski and Sacchi 1987) followed by RT-PCR. This method involves a single acid guanidium thiocyanate-phenol-chloroform RNA extraction followed by an acid phenol extraction and incubation with RNase-free DNase. The sample is subsequently extracted with acid phenol and α₁m-bikunin RNA is amplified with specific primers using a thermocycler (Perkin-Elmer Cetus GeneAmp PCR system 9600; Norwalk, CT) and appropriate reagents (Gene Amp, EZ rTth RNA PCR kit). The α₁m-specific primers were 5'tgctgggagagggcgctaca3' (forward) and 5'gcccatgcagccgccgtact 3' (reverse). The expected 593-bp product was generated by first incubating at 30 min at 60C and 1 min at 94C followed by 40 cycles at 94C for 15 sec and 70C at 30 sec. As a positive control, RTPCR was performed on all tissues using primers specific for glyceraldehyde 3-phosphate dehydrogenase. The product was generated by first incubating at 30 min at 60C and 100 sec at 94C followed by 40 cycles at 94C for 15 sec and 69C at 30 sec. Presence of the predicted 336-bp product ensured that negative results for α₁m-bikunin RNA were not due to RNA degradation in the tissue.

Results

Identification of α₁m in Perfused Rat Tissue

Perfused rat tissues were homogenized and insoluble debris was removed by centrifugation. Aliquots of the supernatants were analyzed by SDS-PAGE, followed by Western blotting using anti-rat α₁m antiserum. The distributions of the different α₁m species in rat plasma and in the perfused rat tissues were compared (Figure 1A). In plasma, several α₁m-immunoreactive bands with apparent molecular masses of 28, 43, 75, 110, 180-210, and 240 kD were evident. These bands specifically reacted with rat α₁m antisera (Figure 1B). The 28-kD band is free uncomplexed rat plasma α₁m and the 240-kD and 180-210 kD bands represent the fibronectin-α₁m and α₁-inhibitor-3-α₁m complexes, respectively (Falkenberg et al. 1990,1994). The 43-kD band has not been characterized. The 240- and 180-210 kD bands were seen in most tissues. It is likely that these also represent fibronectin-α₁m and α-inhibitor-3-α₁m. The 180-210 kD α-inhibitor-3-α₁m complex was absent from the liver, and the 240-kD fibronectin-α₁m complex was absent from the lung and pancreas. As discussed previously, the 110- and 75-kD bands are most likely degradation products of the two complexes (Falkenberg et al. 1990,1994).

Some tissues contained forms of α₁m that were not found in plasma. The liver contained five α₁m isoforms with molecular masses between 40 and 50 kD. The 50-kD α₁m isoform was also present in the kidney, and the heart contained a slightly smaller 45-kD isoform. The pancreas and the eye contained more free α₁m relative to the larger complexes than the other tissues. The 18-kD band observed in the eye homogenate represented a nonspecific reaction because it was present even when the primary antiserum was omitted (not illustrated). The brain contained only the 110-kD band. Western blotting of the same samples using an antibody against a control plasma protein yielded the expected band in plasma, whereas other tissue samples were negative (not shown). This shows that the α₁m bands in the tissues did not originate from plasma contamination.

Figure 1

Identification of α₁m in rat tissues. A rat was perfused and the organs were removed and homogenized in 50 ml of a buffer (100 mM NaCl, 50 mM HEPES, containing 100 mM DCI, 500 mM 1,10-phenanthroline, and 50 mM E-64). The homogenized organs were centrifuged and portions of the supernatants were subjected to nonreducing SDS-PAGE followed by immunoblotting with anti-rat α₁m antiserum (A) or a preimmune antiserum (B). The α₁m related bands identified previously in rat plasma are indicated. The molecular mass markers are indicated in kD.

Immunochemical Localization of α₁m Proteins in Human Tissues

To investigate the tissue distribution of the α₁m proteins, we obtained paraffin-embedded sections of human tissue, including heart, liver, lung, and kidney.

These tissue sections were labeled with antisera to α₁m and detected using a biotin-streptavidin-horseradish peroxidase technique. The studies revealed a granular, cytoplasmic labeling for α₁m in the liver (Figure 2A). This is expected because hepatocytes have been shown to be the primary site of α₁m synthesis (Salier et al. 1987,1993; Lindqvist et al. 1992; Chan et al. 1995). In addition, labeling for α₁m was present in the proximal tubules of the kidney (Figure 2G). This result is also expected because α₁m is known to be filtered by the glomeruli and reabsorbed in the tubules. Immunochemical staining of the heart and lung also demonstrated positive labeling for α₁m, supporting the results of Western blot analysis of these tissues (Figure 1). Specifically, labeling for α₁m was found in the ex-tracellular matrix of blood vessels in the heart (Figure 2B), lung (Figure 2F), liver, and kidney (not illustrated). Furthermore, α₁m was also present on the cell surface of cardiocytes in the heart (Figure 2C) and within the matrix of alveolar septa and airways in the lung (Figures 2E and 2F). No labeling was observed when nonimmune serum was substituted for the primary antibody or when antibodies were absorbed with purified α₁m (Figures 2D and 2H). These results indicate that the labeling was specific for α₁m.

Figure 2

Immunochemical localization of α₁m in human tissues. Sections of human kidney, heart, liver, and lung were labeled with an antibody to α₁m or with nonimmune serum, and the antibody was detected using a biotin-streptavidin-horseradish peroxidase technique. (A) Liver: punctate granular labeling for α₁m is present in the cytoplasm of hepatocytes. (B) Heart: labeling for α₁m is also found in the adventitial matrix of blood vessels in the heart (arrow). (C) Heart: in addition, labeling for α₁m is present on the membrane surface of cardiocytes (brown labeling). (D) Heart: no labeling is present when nonimmune serum is substituted for the primary antibody for α₁m. (E) Lung: α₁m is found in the matrix surrounding airways (arrow). (F) Lung: in addition, labeling for α₁m is present in alveolar septa (arrow) and in the matrix of small blood vessels (^∗). (G) Kidney: immunochemical labeling for α₁m in the kidney is present within the cytoplasm of proximal tubule epithelial cells (brown color). No intracellular labeling is observed in glomeruli (^∗) or in distal tubule epithelium. (H) Kidney: no labeling is present when nonimmune serum is substituted for the primary antibody for α₁m. Bars = 100 μm.

Reverse Transcriptase PCR

The results described above show that α₁m is distributed throughout the body. To determine if the protein was made locally, we examined tissue samples for the presence of α₁m mRNA. Appropriate primers were designed as described above and reverse transcriptase PCR was performed on mRNA extracted from the following human tissues: liver, kidney, lung, heart, spleen, muscle, thyroid, pancreas, and adrenal gland. Figure 3 shows that only the liver and the pancreas produced α₁m RNA. The 0.6-kb PCR band was much weaker in pancreas, suggesting a lower rate of gene transcription and protein synthesis in pancreas than in liver. All tissues contained mRNA for glyceraldehyde 3-phosphate dehydrogenase, indicating that the lack of α₁m mRNA was not due to RNA degradation. These results indicate that α₁m present in the other tissues is not locally synthesized but is produced mainly in the liver, secreted into the blood, and transported to the tissues from the plasma by an unknown mechanism.

Discussion

Previous studies have demonstrated α₁m immunoreactivity in skin, colon, testis, and ovary (Bouic et al. 1984; Odum and Nielsen 1994,1997). We have shown that α₁m is also present in many other tissues, including liver, heart, eye, kidney, brain, lung, pancreas, and skeletal muscle. An intense granular intracellular labeling was observed in hepatocytes, consistent with production and secretion by these cells. Intracellular α₁m labeling was also observed in the proximal tubule cells of the kidney. This was expected because the small size of α₁m promotes uptake, reabsorbance and degradation by the kidneys (Strober and Waldmann 1974; Tejler et al. 1978; Salier et al. 1987,1993; Lindqvist et al. 1992; Chan et al. 1995). Other α₁mcontaining tissues did not show any intracellular labeling but instead demonstrated widespread staining in the interstitial connective tissue matrix. For example, the matrix of blood vessels and the matrix of alveolar septa and airways in the lung were labeled by anti-α₁m. Most of the known plasma proteins, except the immunoglobulins, are synthesized by hepatocytes. In some cases, additional synthesis of plasma proteins takes place in other cell types (Lamontagne et al. 1985; Campbell and Law 1992). The synthesis by these cells is often minor compared to the synthesis by the hepatocytes. α₁m has been reported to be synthesized by the liver (Åkerström and Landin 1985; Salier et al. 1987,1993; Lindqvist et al. 1992; Chan et al. 1995). Other organs, such as kidney in rat and plaice (Karsten et al. 1986; Leaver et al. 1994), stomach in pig (Tavakkol 1991), blood cells in plaice (Leaver et al. 1994), and pancreas in humans (Itoh et al. 1996), have also been suggested to produce α₁m. In the present study, done on human tissue, α₁m mRNA was detected in the liver and, to a lesser extent, in the pancreas, suggesting that human α₁m is mainly produced in the liver, but with a minor contribution from the pancreas. The presence of α₁m in pancreas is presently being investigated by immunohistochemical and biochemical analyses (Lögdberg et al., manuscript in preparation). All other organs in which staining by anti-α₁m was demonstrated lacked α₁m mRNA. The α₁m present in the interstitial connective tissue matrix of different organs is therefore most likely not synthesized locally but is transported to these compartments from blood. This is supported by the fact that although the α₁m-bikunin mRNA content is elevated during inflammation, the total plasma concentrations and urinary excretion of bikunin and α₁m remain unchanged (Falkenberg et al. 1997).

Figure 3

Expression of α₁m. Total RNAs were extracted from the indicated tissues. The extracted RNA was reverse-transcribed and amplified by RT-PCR using primers specific for α₁m mRNA. The PCR products were electrophoresed in a 2% agarose gel and stained with the ethidium bromide. α₁m is secreted by the liver and the 593-bp product was detected as expected. The mRNA was also detected in the pancreas, but all other tissues examined did not appear to contain α₁m mRNA.

α₁m is found in plasma both as a monomer and in complexes with other plasma proteins. These plasma complexes are mainly covalent and have previously been analyzed (Grubb et al. 1983; Falkenberg et al. 1990,1994; Berggård et al. 1997). The biosynthesis and the site of formation of the high molecular weight α₁m complexes are not known. Many low molecular mass plasma proteins are rapidly eliminated from plasma by glomerular filtration. In some cases, the low molecular weight proteins are complex-bound to larger proteins in plasma, resulting in a slower elimination time (Cowan et al. 1990). In accordance with this, it has been speculated that one purpose of the complex formation between α₁m and other plasma proteins may be to prevent loss of α₁m by filtration through the kidney glomeruli (Berggård et al. 1997). The molecular mass of each protein also determines the extent of passive exchange between the intercellular space and plasma. Although larger proteins are less likely to passively penetrate the endothelium of blood vessels, we detected higher levels of high molecular weight α₁m, relative to free α₁m in the tissues. In addition, the brain contained high molecular weight α₁m, although plasma proteins are unable to cross the blood-brain barrier. These findings suggest that the transport from the blood to the tissues is receptormediated.

The results reported here indicate that the biodistribution of α₁m is not restricted to blood. In fact, α₁m is widely distributed in the interstitial matrices and appears to form tissue-specific complexes with other proteins that are not found in plasma. Therefore, α₁m could be regarded as a matrix protein in addition to its previous categorization as a plasma protein. Several possible functions for tissue α₁m can be envisioned. As speculated previously (Odum and Nielsen 1994,1997; Falkenberg et al. 1997), it is possible that the immunosuppressive effects of α₁m, i. e., the inhibition of lymphocyte proliferation and granulocyte chemotaxis and migration, protect “bystander” interstitial tissue from the immune and inflammatory reactions. Furthermore, α₁m is a member of the lipocalin superfamily. Most of the lipocalins are carriers of small hydrophobic prosthetic groups such as retinol, pheromones, odorants, bilirubin, and steroids (Flower 1996). It can be speculated that α₁m is a transporter of such small prosthetic groups from blood to tissues. This would be in agreement with α₁m being synthesized mainly in the liver but found in most organs. Finally, the specific presence of α₁m around blood vessels and in lung airways and alveolar septa resembles that of extracellular superoxide dismutase (EC-SOD), a scavenger of the superoxide anion (Oury et al. 1994). This protein is believed to be important in the defense against superoxide-mediated tissue damage. A similar protective role for α₁m in these domains is possible, as an immunoprotective agent, as a potential scavenger of small toxic substances, or both.